On Mon, May 17, 2010 at 8:37 PM, Mike McGrath <mmcgrath(a)redhat.com> wrote:
On Mon, 17 May 2010, Matt Domsch wrote:
> Now that I've knocked out a few MM bugs, other interesting trends are
> Why are the app* servers running all these in cron.daily? The
> combination of these take approximately 2 hours, and show up very
> noticibly on the collectd graphs. I'm not sure which ones are the
> most IO-intensive for that duration - I haven't caught it during the
> run to watch.
> logrotate - arguably necessary, unless this is done centrally instead?
> makewhatis.cron - we don't have many manpages on the app servers, and
> don't change them that often. Daily? Really?
Well if we were running fedora and running updates daily might not be
fast enough :).
> prelink - we don't start apps that often - they're
If I understand prelink correctly, it really only needs to be run
after an RPM update. We could move this to a monthly cron or a check
that if rpm -V shows prelink problems it is run.
> rkhunter - arguably necessary
> rpm - again, daily? Unclear what this serves us. It's just a file
> listing all the RPMs installed at that moment in time.
> tetex.cron - looks like it'd go fast, so no big deal. But why do we
> have tetex-fonts installed on the app servers?
I think they get pulled in from some dependency chain. Ah yes.. mediawiki
Removing for dependencies:
Remove 16 Package(s)
Reinstall 0 Package(s)
Downgrade 0 Package(s)
> tmpwatch - good to have
> Something to consider revising after the freeze?
I agree with all of the above assessments.
infrastructure mailing list
Stephen J Smoogen.
“The core skill of innovators is error recovery, not failure avoidance.”
Randy Nelson, President of Pixar University.
"We have a strategic plan. It's called doing things.""
— Herb Kelleher, founder Southwest Airlines